Evaluation of Controlled-pore Glass Immobilized lminodiacetate as a Reagent for Automated On-line Matrix Separation for Inductively Coupled Plasma Mass Spectrometry SIMON M. NELMS AND GILLIAN M. GREENWAY" University of Hull Cottingham Road Hull N . Humberside UK .HU6 7RX DAGMAR KOLLER VG Elemental Ion Path Road Three Winsford Cheshire UK C W7 3BX A new iminodiacetate chelating reagent immobilized onto a controlled-pore glass support was evaluated as a substrate for on-line matrix separation for ICP-MS. An automated Fi manifold was constructed incorporating a glass mini-column of the iminodiacetate reagent. One compromise set of optimum conditions was obtained for a range of analytes using the variables of matrix separation flow rate buffer pH and concentration and eluent acid concentration.With a sample volume of 3 ml and an elution volume of 0.3 ml a preconcentration factor of 10 was obtained in addition to matrix separation. The range of elements found to be retained by the column included transition metal cations uranium and lead. These could be quantitatively eluted using nitric acid (0.5 mol dm-3). Calibrations prepared from both pure water and artificial sea-water matrices were found to be comparable in terms of sensitivity illustrating that the presence of a saline matrix did not affect the separation process. Both calibration sets showed good linearity with least squares regression coefficients between 0.996 and 0.999 for the analytes measured. The method gave acceptable reproducibility with precisions (s,) at the 5 ng m1-I level of <5% for 5 replicates.Recoveries between 62 and 113% were obtained for all the elements analysed except Mn which gave a very low recovery (< 35%) under the compromise conditions used. The chelating material was found to have a capacity of approximately 0.1 mmol g-' for a range of elements. The procedure was validated by accurate analysis of the National Research Council of Canada CRMs SLEW-1 (Estuarine Water) and CASS-2 (Coastal Sea- Wa ter). Keywords Flow injection; inductively coupled plasma mtiss spectrometry on-line matrix separation; controlled pore glass; immobilized iminodiacerate ICP-MS has rapidly become one of the most sensitive accurate and reliable trace element measurement techniques. Despite the advantages of ICP-MS its principle disadvantages of low tolerance to dissolved solids (<0.2% m/v) and formation of polyatomic interferences are well known.'*2 In addition the presence of easily ionizable matrix elements such as Nit cause ionization suppression because of the increased electron den- sity in the p l a ~ m a .~ Problems with dissolved solids can be alleviated by diluting the sample but this is unsuitable for samples containing very low levels of analyte species such as sea-waters. Polyatomic interferences cannot be removed by dilution and so degrade the accuracy of quadrupole I(:P-MS * To whom correspondence should be addressed. JkS Journal of Analytical Atomic Spectrometry for isotopes such as 63Cu (due to 40Ar23Na) '9C0 (due to 42Ca'601H) and 7'As (due to 40Ar35C1). The use of high resolution magnetic sector ICP-MS instruments allows mass resolution of the interference from the analyte of intere~t,~ but this can be an expensive solution to the problem. More recently use of the shielded 'cool' plasma has been shown to decrease argon based interferences substantially notably 40Ar'60 on 56Fe by eliminating secondary discharges between the plasma and the sampling interface.' This effect may also reduce matrix-related argon interferences such as 40Ar23Na and 40Ar3'C1 thereby permitting accurate measurements of 63Cu and 75As without the need for matrix separation.The disadvantage here however is that cool plasma conditions reduce the ionization efficiency of elements with ionization potentials greater than 8 eV thereby leading to lower sensi- tivity.Furthermore cool plasma conditions favour the forma- tion of some molecular species such as refractory element oxides and suffer more severe matrix effects in place of argon based interferences.6 To counteract the analytical problems outlined above without resorting to new potentially costly instrumentation considerable effort has been invested in the development of on-line matrix separation using chelating ion- exchange resins. Of the many studies which have been per- formed in this area using a range of detection systems two chelating functional groups namely iminodiacetate (IDA) and 8-hydroxyquinoline (8-HQ) have provoked the most interest. Early on-line matrix separation studies used columns of Chelex-100 resin incorporated into the flow manifold to separate several trace metals from sea-water samples with subsequent detection using AAS7 and ICP-OES.* These studies highlighted that Chelex-100 which contains the IDA ligand covalently bonded onto a polystyrene-divinylbenzene support is difficult to use in on-line systems because of the large volume changes it undergoes when it is converted from a salt form to the acid form.It also requires conditioning steps between each sample analysis resulting in an increase in the analysis time. To address this difficulty alternative more highly cross-linked IDA resins which are less susceptible to dimensional changes have recently been employed in system^.'.'^ These resins are amenable for use in on-line systems but like Chelex-100 require conditioning between samples.An alternative support which is not susceptible to dimensional changes with changing solution composition is controlled-pore glass (CPG). This support possesses a reactive surface which can be easily functionalized with chelating groups and can be conditioned rapidly between samples leading to shorter analysis times. Following the benchmark procedure for covalently bonding 8-HQ to CPG described by Hi1l,I1 numerous successful on-line Journal of .4nalytical Atomic Spectrometry October 1996 Vol. 11 (907-912) 907matrix separation studies using CPG-8-HQ combined with flame AAS,I2,l3 ETAAS14 and ICP-MS15 detection have been described. The location of the separation column in the flow manifold is an important factor in on-line matrix separation studies. In systems developed by Bloxham et a!.' and Beauchemin and Berman,15 in which ICP-MS detection was used the column was incorporated in the flow stream and a separate valve was used to direct the matrix to waste downstream from it.Alternatively it has been suggested that by incorporating the column within a loop across one valve to allow direct switching between the sample and eluent streams the need for a separate additional valve can be circumvented.16 Unfortunately this arrangement allows some matrix to pass into the detector on switching the valve but if the connecting tubing is sufficiently short the residual matrix is kept to an acceptably low level. This design facilitates counter current elution of retained analytes yielding sharper less dispersed peaks and resulting in a reduction in analysis time.In this paper a rapid automated on-line matrix separation system for ICP-MS using an FI manifold is described. The manifold incorporates a mini-column of a new CPG-based IDA resin PROSEP Chelating-1 for the determination of several trace elements in saline samples. This novel material combines the advantages of a CPG support with the efficient chelating performance of the IDA ligand. The immobilized chelate is contained within a glass mini-column located in the loop of an automatic Teflon dual 6 port valve. The retained analytes are eluted counter current to the sample flow to yield improved less dispersed peaks. The system described is based on fixed volume injection rather than time based sampling and gives a 10-fold preconcentration in addition to the required matrix separation for a 3 ml sample volume.Data from the transient eluted peaks was collected using the peak jumping data acquisition mode of the instrument with a suitable uptake delay to allow the eluted sample to travel from the column to the plasma. A full optimization of the system is described including recovery and capacity results. Validation of the procedure using National Research Council of Canada (NRCC) CASS-2 (Coastal Sea-water) and SLEW-1 CRMs (Estuarine water) is also presented. EXPERIMENTAL Reagents The iminodiacetate reagent (PROSEP Chelating- 1 Bioprocessing Consett Co. Durham UK) was used as sup- plied. The preparation and structure of this material cannot be described here for commercial reasons. The chelating mate- rial (0.04 g) was packed as a dry powder into a glass mini- column (2.5 cm x 3 mm Omnifit Cambridge UK).High-purity deionized water (18 MR cm resistivity Elgastat UHQ PS Elga High Wycombe Buckinghamshire UK) was used throughout. Elemental stock solutions (1000 pg ml- ' SpectrosoL Merck Poole Dorset UK) were used in the preparation of calibration solutions. Ammonium acetate buffer (Sigma Poole Dorset UK) was prepared from the solid and purified on-line by passing through a column of Chelex-100 (Sigma). Adjustments in pH were made using glacial acetic acid or aqueous ammonia solution as appropriate. Artificial sea-water (Instant Ocean Aquarium Systems Mentor OH USA) was prepared by dissolving approximately 330g of the powder in 101 of water. Samples of the NRCC (Ottawa Canada) CASS-2 and SLEW-1 were introduced into the mani- fold as supplied.Instrumentation ICP-MS measurements were made using a VG Elemental PlasmaQuad 2 Plus (VG Elemental Winsford Cheshire UK). The instrument was calibrated and optimized prior to oper- ation using a solution containing the elements Be Mg Co Y La Eu and Bi at 10 ngml-' in a matrix of 2% nitric acid. The transient analyte peaks were monitored in the peak jumping mode using the data acquisition and instrument operating parameters given in Table 1. The automated matrix separation procedure was carried out using the Preplab unit (VG Elemental). For the PROSEP Chelating-1 capacity evalu- ation measurements were made using a Perkin-Elmer Plasma 40 ICP-OES instrument ( Perkin Elmer Beaconsfield Buckinghamshire UK) the operating parameters of which are given in Table 1.Matrix separation manifold The automated matrix separation manifold is illustrated in Fig. 1. The sample loop (3 ml) was located across the front of the dual 6 port Teflon injection valve (D-6-V) on the Preplab unit. The PROSEP Chelating-1 material was contained within a glass mini-column (2.5 cm x 3 mm id Omnifit Cambridge UK) incorporated in a loop across the rear of the D-6-V. All the manifold connections were made using 0.8 mm id PTFE tubing. The reagents were pumped using the two peristaltic pumps on the Preplab. With the configuration used the matrix separation flow rate could be varied but the eluent flow rate was fixed at 1.5ml min-'. Using the two valves supplied on the Preplab both the sample uptake line and on-line buffer streams could be switched to water to rinse the sample loop between samples and the column after sample loading. Table 1 Operating parameters for the ICP and ICP-MS instruments ICP instrument - Aerosol gas flow rate/l min- ' Intermediate gas flow rate/l min-' Outer gas flow/l min-' Nebulizer Cross flow Spray chamber Ryton ICP emission wavelengths - Element V Cr Mn Fe c o Ni c u Zn Cd Ce Pb U ICP-MS instrument - Forward power/W Aerosol gas flow rate/l min-' Outer gas flow/l min-' Reflected power/W Intermediate gas flow rate/l min-' Spray chamber Glass water cooled 10 "C Nebulizer de Galan type Peak jumping acquisition parameters - Points per peak Dwell time Detector mode Selected isotopes (ICP-MS) - 48Ti 49Ti 51V "Mn "Co 60Ni 63Cu 64Zn 65Cu 114Cd 208pb 2 3 8 ~ 9 x0.75 0.6 12 Wavelength/nm 309.311 205.552 257.610 238.204 238.892 221.647 324.754 213.856 214.438 413.765 220.35 3 385.958 1350 0.939 13.0 0 1 .o 3 10.24 ms Pulse counting 908 Journal of Analytical Atomic Spectrometry October 1996 1/01.1 1SAMPLE WATER Fig. 1 Flow injection manifold Matrix Separation Procedure The automated matrix separation procedure began by opening the D-6-V and valves 1 and 2 (V1 and V2) then starting peristaltic pump 2 (P2). This allowed sample to pass into the sample loop and also eluent acid to pass as a continuous stream through the chelating column. After a period of 50s peristaltic pump 1 (Pl) was started to pump buffer through to the D-6-V via an on-line purification column containing Chelex-100 (A on Fig.1). After a further 30 s buffer had reached the D-6-V and the sample loop was full. At this point V1 closed followed immediately by the D-6-V. Closing V1 stopped the sample uptake by switching to water and c1,osing the D-6-V allowed the sample loop contents to be eluted by water the eluent was then mixed with the buffer (at point B) and finally passed through the chelating column. In this way trace elements in the buffered sample were retained on the column and the matrix and buffer species were passed to waste. Once the sample had passed into the column V2 swiwhed from buffer to water to allow residual buffer and matrix species in the system to be rinsed to waste. After this rinse period the D-6-V and V1 were switched back to open.This allowed acid back through the column thereby eluting the retained analytes into the ICP-MS and simultaneously enabled a new sample to load during the ICP-MS analysis period. As the ICP-MS analysis time was similar to the initial 50s sample loading period the latter could be skipped for subsequent repeats on the same sample by incorporation of a loop command in the Preplab program thereby decreasing the total analysis time. for each element was evaluated using eqn. 1 Determination of Exchange Capacity and Recovery The capacity of PROSEP Chelating-1 was determined for a range of elements using a batch method. Solutions of the selected analytes (200 pg ml- ') were prepared in artificial sea-water (10 ml) then added to ammonium acetate buffer (1.5 mol dm-3 10 ml).Each solution was adjusted to pH 6.5 (pH 8 for Cr and Mn) and added to 0.05 g of the dry chelating material. The mixtures were shaken and left to equilibrate overnight. The reduced concentration of each analyte i n the supernatant solution was then measured uersus the original concentration using ICP-OES. The capacity of the material where C is the capacity (mmol g-') ci and cf are the concen- trations (pg ml-') of the element before and after equilibrium M is the relative atomic mass (g mol-') of the element and u is the volume (ml) of solution equilibrated with a mass m (g) of PROSEP Chelating-1. Recoveries for a range of elements were evaluated dynamically for both a fresh and an aged column of PROSEP Chelating-1 by passing an artificial sea- water solution spiked with the selected analytes at 10 ng ml-' through the automated FI manifold.The aged column had been in use for approximately 240 h before the recovery study. A manual valve (Rheodyne 5020 Supelco Poole Dorset UK) fitted with a 0.3 ml loop was incorporated into the eluent acid stream of the manifold to yield a 10-fold preconcentration on elution. Using this configuration the eluent acid instead of being pumped continuously was injected as a known fixed volume via the manual valve into a water carrier stream. Recoveries were evaluated by comparing samples eluted from the column with a 100 ng ml-' solution prepared in the eluent matrix and injected via the manual valve. RESULTS AND DISCUSSION Optimization of the Matrix Separation Procedure The procedure was optimized with respect to the matrix separation flow rate buffer pH buffer concentration and eluent acid concentration using a univariate approach.For the optimization procedure samples were prepared by spiking a synthetic sea-water solution with the selected analytes (10 ng ml-') followed by acidification with O.lml of concen- trated nitric acid to 100 ml of sample. Acidification was performed to prevent trace element loss from solution by adsorption or precipitation. The optimum conditions for on-line matrix separation were 1.5 mol dm-3 of ammonium acetate (pH 6.5); 0.5 mol dm-3 of nitric acid; matrix separation flow rate 5.0 ml min-'; elution flow rate 1.5 ml min-'; and total analysis time 5 min. The effect of flow rate on the matrix separation procedure is illustrated for selected elements in Journal of Analytical Atomic Spectrometry October 1996 Vol.11 909Fig. 2. Up to a flow rate of 5 ml min-l no significant change in response was observed for the elements analysed. Faster flow rates than 5 ml min-' introduced back pressure problems and slower rates led to increased analysis time. Therefore 5 ml min-l was selected for the rest of this study. The response of the column to selected analytes with changing buffer pH is illustrated in Fig. 3. Since the optimum buffer pH for matrix separation varies between elements a compromise pH must be selected for multi-element analysis. On the basis of the element responses shown in Fig. 3 pH 6.5 was chosen as values below this gave decreased retention of some analytes.Values above this level reduced the capacity of the acetate buffer and also lead to increased retention of Ca as the PROSEP Chelating-1 has some affinity for this element. The effect of the ammonium acetate buffer concentration on the retention of selected elements is illustrated in Fig. 4. The results show that the buffering process is effective for many elements at a concen- tration of 0.05 mol dm-3. However ionogenic retention of Ca at this level is high because the buffer cation concentration is insufficient to displace this element completely. Retention of 48Ca interferes with 48Ti and at zero buffer concentration 63Cu and 'lV are also interfered with because of the formation of 40Ar23Na and 35Cl'60 in the plasma with residual Na and C1 respectively. It was found that a buffer concentration of 1.5 mol dm-3 was required to remove the Ca problem.This high concentration did not affect retention of the selected analytes except Mn and U (as UOzf). These two elements show similar behaviour to Ca on PROSEP Chelating-1 which is expected from the relative positions of these three metals in the Irving-Williams series.17 The effect on analyte elution with increasing eluent acid concentration is illustrated for selected I 2.0 2.5 3.0 3.5 4.0 4.5 5.0 FIOW rate/ml min-' Fig. 2 Effect of matrix separation flow rate. A Mn; B Ni; C Cu; D U and E Zn 30 25 20 15 10 5 0 4.0 4.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0 PH Fig.3 D Ni; and E U Effect of ammonium acetate buffer pH. A V; B Mn; C Co; 60 5 50 * 40 2 30 (d Y a v) 4- 8 g 20 10 0 h D 0 0.5 1 1.5 2 [Buffer]/mol dm-3 Fig.4 Co; C Ni; D U; and E Ca Effect of ammonium acetate buffer concentration. A Mn; B analytes in Fig. 5. In this study quantitative elution was observed for all the elements investigated at an acid concen- tration of 0.5 mol dm-3. Acid concentrations greater than 2.0 mol dmP3 were not investigated as these could degrade the column. To determine the! effect on the column of repeated use a second column was prepared from the same batch of PROSEP Chelating-1 and the two columns were compared in terms of element recoveries (Table 2). Recoveries between 62 and 113% were obtained except for Mn for which the recovery was only 30%. This was due to the relatively low affinity of the resin for Mn under the matrix separation conditions used. The results showed that there was a slight reduction in the column performance after use for approximately 240 h.Capacities of the PROSEP Chelating-1 material were evaluated for a wide range of elements (Table 3) and found to be lower (by typically 66%) than those of the polymeric based IDA resin Chelex-100.'8 This is expected as the rigid CPG support cannot expand to allow greater access to the complexation I 0 0.5 1 .o 1.5 2.0 [Acid]/mol dmm3 Fig. 5 D Cu; and E U Effect of nitric acid eluent concentration. A V; B Mn; C Ni Table 2 Recovery comparison between aged and fresh column Recovery (YO) Element Ti V Mn c o Ni c u U Aged column 76 too 30 98 105 71 62 Fresh column 82 101 31 113 110 97 61 91 0 Journal of Analytical Atomic Spectrometry October 1996 Vol. 11Table 3 Capacity results for PROSEP Chelating-1 Element V Cr Mn Fe co Ni c u Zn Cd Ce Pb U Capacity/mmol g- ' 0.22 0.10 0.14 0.29 0.13 0.11 0.1 1 0.13 0.10 0.10 0.13 0.04 sites.Nonetheless the capacity of PROSEP Chelating-1 I S still more than sufficent for use in trace element studies. Evaluation of the Effect of Residual Matrix Species In a previous paper," a procedure for evaluating the influence of residual matrix on the ICP-MS results using the 63Cu 65Cu isotope ratio was described. This procedure was based on the observation that the 63Cu 65Cu isotope ratio is anomalously high if Na is present in the plasma due to polyatomic overlap of 40Ar23Na on 63Cu. However this method of identifjing a residual matrix problem does not cover the possible inter- ference of 33S'602+ on 65Cu which could yield a low 63Cu 65Cu ratio.In this paper this potential inaccuracy has been addressed by presenting data for the 48Ti:4 9Ti ratio in addition to 63Cu 65Cu data to clarify the efficiency of the matrix separation procedure. The 48Ti :"Ti ratio clearly identifies the effect of residual Ca on the process by virtue of the 48Ca/48Ti isobaric overlap. Like 65Cu 49Ti can be affected by a sulfur interference 33S160 but since 33S has a natural abundance of only 0.75% all 33S based interferences will be low. Furthermore since the Table 4 pure saline and matrix separated water samples Isotope ratio measurements for 63Cu 65Cu and 48Ti 4"Ti for Sample description 63Cu 65Cu ratio* 48Ti 49Ti ratio* value 300 pl injected ( 10 ng ml- ') Natural ratio accepted 2.24 13.42 Sea-water (10 ng mi-') 5.55 f 3.07 64.04 & 5.13 0.5 mol dm-3 HNO 2.22 f 0.10 13.45k0.57 injected Matrix separation from 2.21 f 0.05 13.42 f 0.88 spiked pure water ( 5 ng m1-l) spiked sea-water ( 5 ng mi-') Matrix separation from 2.17 f 0.12 13.45 & 0.47 * Values quoted with a range of 2 s (n = 5 ) except natural ratios.Table 5 Comparison between spiked pure and sea-water calibrations sulfate present in the sea-water samples is not retained by the column this source of interference should be negligible. Table 4 illustrates that residual matrix does not affect the isotope ratio measurement for 63Cu 65Cu or 48Ti 49Ti thereby illustrating that the matrix separation procedure is effective with respect to both Na and Ca removal.The results presented for an injected artificial sea-water sample (Table 4) clearly show a severe increase for both the apparent 63Cu 65Cu and 48Ti 49Ti ratios illustrating that FI although effective in reducing block- age problems with high salt samples cannot remove the associated polyatomic interferences. Matrix separation is there- fore essential if accurate results are to be obtained. Calibration and Analysis of Certified Reference Materials Using ICP-MS linear calibrations were obtained over the range 0 to 10 ng ml-' for analytes in both spiked pure water and spiked synthetic sea water matrices as described in Table 5. The two calibration sets compared well therefore validat- ing the use of simple pure water calibration solutions for quantifying analytes in more complex matrices.The matrix separation procedure was validated by analysis of the two certified reference materials SLEW-1 and CASS-2. The analytes were quantified by external calibration against acidified multi-element (Mn Cu Zn Ni Co Cd) pure water standards processed through the manifold. Three repeat analy- ses were made at each concentration and for the reference materials. The calibrations generally showed good linearity with least squares regression coefficients of 0.997 to 0.999 being obtained across the concentration ranges 0-20 ng ml-' (Mn) 0-4 ng ml-' (Cu Ni Zn) and 0-0.1 ng ml-' (Cd Co). Precision (measured as RSD) for the selected analytes were in the range 0.5-5.5 O/O. Results for the certified reference materials analysis are given in Table 6.For both materials good agreement between the found and certified values was obtained for all the elements measured except for Mn. A consistently low result was obtained for this element in both SLEW-1 and CASS-2. This was interpreted to be a consequence of the low affinity of PROSEP Chelating-1 for Mn under the matrix separation conditions used (see Figs. 3 and 4). Direct analysis Table 6 Analysis results for the certified reference materials SLEW-1 and CASS-2 SLEW-1 CASS-2 Element Found* Certified* Mn 7.97k0.72 13.1 f0.8 Co 0.040 f 0.003 0.046 f 0.007 Ni 0.75 1 f 0.074 0.743 f 0.078 c u 1.72 0.30 1.76 f 0.09 Zn 0.74f0.03 0.8650.15 Cd 0.01 5 f 0.002 0.01 8 2 0.003 Found* Certified* 1.84 Ifr 0.07 1.99 k 0.15 0.028 & 0.001 0.025 f 0.006 0.264 & 0.006 0.298 f 0.036 0.704 f 0.10 0.675 k 0.039 1.95 f 0.05 1.97 k0.12 0.017f0.002 0.019f0.004 * Concentrations in ng ml-'.Uncertainties expressed as 2 s of the instrument response to each analyte (95% confidence limit n= 3). Culibrations front spiked pure water - Parameter RSD (YO) at 5 ngml-' ( n = 5 ) Correlation coefficient r Sensitivity ( counts ng-') Detection limit/ng ml- ' ( 5 s n = 5 ) Calibrations from synthetic sea-water- Parameter RSD (YO) at 5 ngml-' (n=5) Correlation coefficient r Sensitivity ( counts ng-' ml) Detection limit/ng ml - 5 s n = 5 ) 55Mn 4.6 0.9997 1.28 0.10 55Mn 3.2 0.9987 1.12 0.56 5 9 c ~ 4.0 0.9998 4.57 0.02 59c~ 2.4 0.9997 5.14 0.0 1 63cu 2.4 0.9990 2.3 1 0.05 63cu 2.3 0.9999 2.58 0.09 64Zn 3.3 0.9989 1.27 0.20 64Zn 1.1 0.9956 1.35 0.44 '14Cd 2.2 0.9998 1.75 0.09 ' 14Cd 2.0 0.9996 1.77 0.07 .Journal of Analytical Atomic Spectrometry October 1996 Vol.11 91 1of the reference materials for comparison with the matrix separation procedure could not be performed because this led to cone and injector blockage and signal suppression in the plasma. Injection of saline samples without prior matrix separa- tion circumvented blockage problems but significant polya- tomic interferences and signal suppression remained. These problems coupled with the low analyte concentrations present made direct injection impractical for the reference material analysis. CONCLUSIONS Batch capacity measurements for a range of elements showed that PROSEP Chelating-1 was comparable and in some cases superior to CPG-8-hydroxyquinoline materials but lower than polymeric based chelators due to the inflexibility of the CPG support.However this rigidity combined with the good chemical stability physical robustness rapid surface reactivity and wide elemental application of PROSEP Chelating- 1 made the material a highly effective reagent for the matrix separation procedure undertaken in this study. A second paper discussing the effects of elemental and humic acid interferences as well as the application of the system to real river water and effluent samples is currently in preparation. S. M. N would like to thank the EPSRC and VG Elemental for their provision of funding and equipment for the project Peter Clarke of Bioprocessing Ltd. for supplying the PROSEP Chelating-1 material and Sarah Dolman for producing Fig.1. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Beauchemin D. TrAC Trends Anal. Chem. (Pers. Ed). 1991 10 71. McLaren J. W. At Spectrosc. 1993 14 191. Houk R. S. and Olivares J. A. Anal. Chem. 1986 58 20. Reed N. M. Cairns R. O. Hutton R. C. and Takaku Y. J. Anal. At. Spectrom. 1994 9 881. Tanner S. D. Paul M. Beres S. A. and Denoyer E. R. At. Spectrosc. 1995 16 16. Tanner S. D. J. Anal. At. Spectrom. 1995 10 905. Olsen. S. Pessenda L. C. R. Ruzicka J. and Hansen E. H. Analyst 1983 108 905. Hartenstein S. D. Ruzicka J. and Christian G. D. Anal. Chem. 1985 57 21. Bloxham M. J. Hill S. J. and Worsfold P. J. J. Anal. At. Spectrom. 1994 9 935. Taylor D. B. Kingston H. M. Nogay D. J. Koller D. and Hutton R. C. J. Anal. At. Spectrom. 1996 11 187. Hill J. M. J. Chromatogr. 1973 76 455. Malamas F. Bengtsson M. and Johansson G. Anal. Chim. Acta 1984 160 1. Fang Z. and Welz B. J. Anal. At. Spectrom. 1989 4 543. Nakashima S. Sturgeon R. E. Willie S. N. and Berman S. S. Fresenius’ 2. Anal. Chem. 1988 330 592. Beauchemin D. and Berman S. S. Anal. Chem. 1989,61 1857. Fang Z. in Flow Injection Separation and Preconcentration VCH Weinheim 1993 Ch. 4. Shriver D. F. Atkins P. W. and Langford C. H. in Inorganic Chemistry Oxford University Press 1990 Ch. 7 p. 222. Bio-Rad Laboratories Chelex-100 Instruction Manual 1996. Nelms S. M. Greenway G. M. and Hutton R. C. J. Anal. Atom. Spectrom. 1995 10 929. Paper 61041 01 I Received June 11 1996 Accepted August 13 1996 91 2 Journal of Analytical Atomic Spectrometry October 1996 Vol. 11